Prefix delegation is used to assign a network address prefix to a host such as a user site, and is described in S. Miyakawa, R. Droms, “Requirements for IPv6 Prefix Delegation”, RFC 3769, June 2004. When configuring an IP network, IP prefixes are delegated down the network hierarchy using manual methods. It would be advantageous to automated prefix delegation in order to reduce network management costs, and also to support network mobility by allowing for automatic delegation of prefixes to a moving network that attaches to a parent network. Automated prefix delegation technology has not been deployed widely yet, even though a number of proposals have been brought forward, for example in US 2003/0182445 A1, WO 2007/149025 A1 and US 2006/0013193 A1. However, the mechanisms described in those proposals do not address the problem of arbitrary edge network topologies with many hierarchical levels.
Consider, for example, a network N levels down in a hierarchy of edge networks, where each network is attached to M parent networks. This network has MN different paths to the core network. Potentially, a network can have as many prefixes as there are different paths to the core network, since prefixes can be delegated along each of the paths. However, each network only needs a few of the MN paths, the exact number depending on the degree of multi-homing that is needed in order to provide sufficient redundancy to ensure that network can be reached in the event of failure of other networks on one of the paths to the network. There is therefore a need for criteria and mechanisms to select the best paths and the associated prefixes in order to reach the network. The best path is defined according to a specific policy routing criterion such as the QoS, the capabilities, or the cost of the path. It should be noted that since each prefix is associated with a specific path to the core network, the prefix selection implies a routing decision.
Prefix selection criteria based on the nearest prefix delegation server, or round-trip time to this server, are described in US 2003/0182445 A1. However, these criteria are not sufficient for the automated assignment and selection of prefixes across the edge topology based on general policy routing rules such as QoS, security, or cost. US 2003/0182445 A1 does not include a mechanism for coupling between prefix delegation and policy routing. In fact, policy routing is largely beyond the scope of US 2003/0182445 A1. Moreover, US 2003/0182445 A1 is based on a centralized prefix server, which is a single point of failure. In addition, there are scalability issues associated with a centralized prefix server, since it may need a capacity upgrade as the network grows.